PSI - Issue 28

Luigi Mario Viespoli et al. / Procedia Structural Integrity 28 (2020) 344–351 Author name / Structural Integrity Procedia 00 (2019) 000–000

346

3

curvature of the sheathing was kept, in order to increase buckling resistance and to avoid the introduction of additional forming of the sheath, obtaining specimens corresponding to the geometry in figure 1. The tests were performed in tensile- compression loading mode which corresponds to the loading mode in the sheathing when the cable is subjected to bending. All tests were conducted in air at room temperature with a Zwick/Roell LTM Electrodynamic machine equipped with a 10 kN load cell. Digital image correlation (DIC) technique was used to control the longitudinal strain range according to the procedure schematically reported in figure 2: the test was started setting the positions values calculated based on tensile properties of the material at the target strain rate. During the first cycles, DIC images were immediately post-processed to obtain the actual strain range values from the surface of the specimen. If the values obtained differed from the target strain range (with a tolerance of about 5%), the clamp position was adjusted and the control repeated. It is assumed that small adjustments done in the initial part of the specimen’s fatigue life have negligible effect on the result. This correction of the imposed displacement was necessary due to the highly plastic behaviour of the material to account for the unavoidable clamping pressure. The use of DIC is preferred to the use of an extensometer because of the absence of contact with the specimen during testing which may create concentrators on the specimen’s surface which will drive the fatigue crack initiation. However, it was not possible to execute real time DIC post processing and control the machine in closed loop without manual intervention. The tests were performed in strain control because of the plastic behaviour even at very low strain ranges. Plasticity is increased by a reduction of strain rate. The same characteristic would lead to the material permanently deforming and settling to a new length if a displacement ratio different from -1 was to be used. A total of 23 specimens were tested, 9 at 1E-3 s 1 and 14 at 1E-2 s-1, the results of which are reported in figure 4 in terms of cycles to failure vs longitudinal strain range, as computed by DIC in the central part of the specimen. Even if no strong stress-strain difference was detected in the tensile testing, time dependent damage is active and shows a different influence according to the strain rate: the increased duration of a cycle at 1E-3 s-1 compared to 1E-2 s-1 makes so that the former presents a reduced fatigue life than the latter in terms of cycles to failure. The observation of the evolution of the stress range in the initial part of the testing shows a negligible relaxation for the specimens tested at 1E-2 s-1, while a reduction of the stress range of more than 10 % happened only for two of the specimens tested at 1E-3 s-1, with the others showing limited relaxation as well. The difference detected is quantified in an increase of strain range from 2.46 % to 6.12 %. considering a 50 % failure probability at 2E-6 cycles. The reduction of resistance in terms of strain range is not constant, but inversely proportional to the load level: the fatigue curve slope is indeed 0.558 for 1E-3 s-1 and 0.346 for 1E-2 s-1 over the tested range.

Fig. 1. Cable section scheme and fatigue specimen geometry. Longitudinal direction of cable and specimen correspond.